https://github.com/bcbeidel/dos
Experimental AI skills library for data work
https://github.com/bcbeidel/dos
Last synced: 3 months ago
JSON representation
Experimental AI skills library for data work
- Host: GitHub
- URL: https://github.com/bcbeidel/dos
- Owner: bcbeidel
- License: mit
- Created: 2026-03-22T00:02:29.000Z (3 months ago)
- Default Branch: main
- Last Pushed: 2026-03-22T13:52:08.000Z (3 months ago)
- Last Synced: 2026-03-22T18:07:35.841Z (3 months ago)
- Size: 71.3 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
- Agents: AGENTS.md
Awesome Lists containing this project
README
# dos
> **Experimental** — this repository is a work in progress and subject to breaking changes.
A Claude Code plugin housing skills for facilitating data engineering work.
## Installation
```bash
claude plugin install dos
```
Or for local development:
```bash
claude --plugin-dir ./dos
```
## Skills
Skills are namespaced under `dos:` and invocable as `/dos:`.
| Skill | Description |
|-------|-------------|
| `dos:evaluate-source` | Assess a data source's technical characteristics before pipeline construction. |
| `dos:scope-data-product` | Define what a data product needs to be, driven by consumption intent. |
| `dos:select-model` | Choose a data modeling approach (Kimball, Data Vault, OBT) based on constraints. |
| `dos:define-contract` | Define or update an ODCS v3.1-aligned data contract for a data product. |
| `dos:assess-quality` | Set up quality engineering with dimensions, scoring, and validation tooling. |
| `dos:design-pipeline` | Architecture a data pipeline from source to serving layer. |
| `dos:implement-source` | Generate dlt pipeline code and dbt source definitions from data product artifacts. |
| `dos:implement-models` | Generate dbt models, schema YAMLs, tests, and contract enforcement from data product artifacts. |
| `dos:review-pipeline` | Audit an existing data pipeline against best practices. |
## Workflow
Skills chain together — each produces artifacts that downstream skills consume. Start anywhere; skills work independently but are most effective in sequence.
```
Discover Scope Design Build Verify
┌────────────┐ ┌─────────────┐ ┌───────────────┐ ┌──────────────┐ ┌─────────────┐
│ evaluate- │─▶│ scope-data- │─▶│ select-model │ │ implement- │ │ review- │
│ source │ │ product │ │ define- │─▶│ source (EL) │─▶│ pipeline │
│ │ │ │─▶│ contract │ │ │ │ │
│ │ │ │ │ assess- │ ├──────────────┤ │ │
│ │ │ │─▶│ quality │─▶│ implement- │─▶│ │
│ │ │ │ │ design- │ │ models (T) │ │ │
│ │ │ │─▶│ pipeline │─▶│ │ │ │
└────────────┘ └─────────────┘ └───────────────┘ └──────────────┘ └─────────────┘
```
**Typical flow:**
1. `/dos:evaluate-source` — assess a source system (e.g., `postgres-orders-db`)
2. `/dos:scope-data-product` — define what the data product needs to be
3. `/dos:define-contract` + `/dos:assess-quality` + `/dos:design-pipeline` — specify the contract, quality rules, and architecture (any order)
4. `/dos:implement-source` — generate dlt pipeline + dbt source YAML
5. `/dos:implement-models` — generate dbt models, tests, and contract enforcement
6. `/dos:review-pipeline` — audit the result; findings loop back to upstream skills
Each skill checks for existing artifacts and adjusts its workflow accordingly. You don't have to start from step 1 — jump in wherever your project is.
## Feedback
After using a skill, [file feedback](../../issues/new?template=skill-feedback.yml) to help improve it. Report what worked, what didn't, and suggestions.
## Development
Skills live in `skills//SKILL.md`. Each skill directory may also contain supporting scripts and reference docs.